1. © Employee Benefit Research Institute 2016 Evidence from the Latest Research on Consumer-Directed Health Plans Mid-America Coalition on Health Care May 19, 2016 Paul Fronstin, Ph.D. Director, Health Research and Education Program Employee Benefit Research Institute Washington, DC Copyright© - Employee Benefit Research Institute Education and Research Fund, 1978-2016. All rights reserved. The information contained herein is not to be construed as an attempt to provide legal, accounting, actuarial, or other such professional advice. Permission to copy or print a personal use copy of this material is hereby granted and brief quotations for the purposes of news reporting and education are permitted. Otherwise, no part of this material may be used or reproduced without permission in writing from EBRI-ERF.

2. © Employee Benefit Research Institute 2016 Percent of Private or Employment-Based Health Insurance Market Enrolled in CDHP, 2015 2 Note: AHIP & NHIS estimates are HSA only. EBRI, KFF, & Mercer estimates are HSA & HRA.

3. © Employee Benefit Research Institute 2016 HSA & HRA Offer Rates, 2005-2015 3 Source: Kaiser Family Foundation.

4. © Employee Benefit Research Institute 2016 Percentage of Employers Offering an HSA or HRA, by Firm Size, 2015 4 Source: Kaiser Family Foundation.

5. © Employee Benefit Research Institute 2016 HSA/HRA Offer Rates, by Firm Size, 2011-2015 & 2018 Projections 5 Source: Mercer’s National Survey of Employer-Sponsored Health Plans.

6. © Employee Benefit Research Institute 2016 HSA/HRA as the “Only Plan Option” in Employment- Based Market on the Rise, 2012-2016 (aka Full Replacement Plan) 6 Source: Willis Towers Watson/National Business Group on Health, 2015 High- Performance Insights: Best Practices in Health Care.

7. © Employee Benefit Research Institute 2016 HSA Assets Reached $30.2 billion in 2015: 16.7 million Accounts 7 Source: Devenir.

8. © Employee Benefit Research Institute 2016 EBRI/Greenwald & Associates Consumer Engagement in Health Care Survey The EBRI/Greenwald & Associates Consumer Engagement in Healthcare Survey was conducted online using a panel. Oversample records were pre-screened using an online omnibus study. Eligibility: Americans age 21 to 64 with private health insurance coverage Sample Groups:1.Traditional samplen=1,490 2.CDHP sample*n=1,285 (259– national; 1,026 – oversample) 3.HDHP sample*n=815 (255 – national; 560 – oversample) *Groups 2 and 3 are composed of national sample records and oversample records Weighting:The national sample is weighted by gender, age, income, ethnicity, education and region to reflect the actual proportions in the population. The CDHP and HDHP samples are weighted by gender, age, income, and ethnicity. Survey Dates:August 4th – August 21th, 2015 Survey Length:12 minutes (mean) Response rate:34.4% 8

9. © Employee Benefit Research Institute 2016 Cost-Conscious Decision Making, by Type of Health Plan, 2015 9 Source: EBRI/Greenwald & Associates Consumer Engagement in Health Care Survey, 2015.

10. © Employee Benefit Research Institute 2016 Informed Decision Making for Health Plan Choice, by Type of Health Plan, 2015 10 Source: EBRI/Greenwald & Associates Consumer Engagement in Health Care Survey, 2015.

11. © Employee Benefit Research Institute 2016 Individual Participates in Wellness Program Offered by Employer, Among Those Offered a Wellness Program, by Type of Health Plan, 2015 11 Source: EBRI/Greenwald & Associates Consumer Engagement in Health Care Survey, 2015.

12. © Employee Benefit Research Institute 2016 Self-Reported Health Status, by Plan Type, 2015 12 Source: EBRI/Greenwald & Associates Consumer Engagement in Health Care Survey, 2015.

13. © Employee Benefit Research Institute 2016 Percentage Currently Smoking Cigarettes, by Plan Type, 2015 13 Source: EBRI/Greenwald & Associates Consumer Engagement in Health Care Survey, 2015.

14. © Employee Benefit Research Institute 2016 CDHP Enrollees Older than Traditional Plan Enrollees, 2015 14 Source: EBRI/Greenwald & Associates Consumer Engagement in Health Care Survey, 2015.

15. © Employee Benefit Research Institute 2016 CDHP Enrollees Higher Income than Traditional Plan Enrollees, 2015 15 Source: EBRI/Greenwald & Associates Consumer Engagement in Health Care Survey, 2015.

16. © Employee Benefit Research Institute 2016 CDHP Enrollees More Educated than Traditional Plan Enrollees, 2015 16 Source: EBRI/Greenwald & Associates Consumer Engagement in Health Care Survey, 2015.

17. EBRIs Center for Research on Health Benefits Innovation Helping employers assess the impact of plan design —with the goal of increasing consumer engagement— on cost, quality, and access to health care 17

18. Research Partners American Express Ameriprise Aon Hewitt Blue Cross Blue Shield Association Boeing Deseret Mutual Federal Reserve Employee Benefits System General Mills Healthways IBM JP Morgan Chase Kaiser Permanente Mercer Pfizer 18

19. © Employee Benefit Research Institute 2016 Findings from the Longest-Ever Study of a Full- Replacement HSA-Eligible Health Plan A large Midwestern employer replaced PPOs with an HSA-eligible plan on Jan. 1, 2007 Choice of two deductibles: $1,250 (individual)/$2,500 (family) $2,500 (individual)/$4,300 (family) Pharmacy and medical administrative claims data and insurance enrollment information obtained from a large employer Data used from Jan. 1, 2006 – Dec. 31, 2010 Deductibles and HSA contributions unchanged over 5-year period Between 10,000 & 18,000 continuously enrolled during the 5-year period Data from second employer used to create a comparison group 19

20. © Employee Benefit Research Institute 2016 Topics Addressed "Consumer-Directed Health Plans Reduce The Long-Term Use Of Outpatient Physician Visits And Prescription Drugs." Health Affairs, June 2013. "Health Care Spending after Adopting a Full Replacement, High-Deductible Health Plan With a Health Savings Account: A Five-Year Study." EBRI Issue Brief, July 2013. "Medication Utilization and Adherence in a Health Savings Account-Eligible Plan." American Journal of Managed Care, December 2013. "Brand-Name and Generic Prescription Drug Use After Adoption of a Full-Replacement, Consumer-Directed Health Plan With a Health Savings Account.“ EBRI Notes, March 2014. "Quality of Health Care After Adopting a Full-Replacement, High-Deductible Health Plan With a Health Savings Account: A Five-Year Study.“ EBRI Issue Brief, September 2014. 20

21. © Employee Benefit Research Institute 2016 Summary of Findings – Use of Health Care Services Outpatient office visits fell in all years Prescription drug fills fell in all years Both generic and brand name use fell Brand name fell more than generics, so generic dispensing rate increased Medication adherence fell for hypertension, dyslipidemia and diabetes in year 1 May be rebounding in year 2 for dyslipidemia and diabetes No effect on Asthma/COPD or depression Emergency department visits higher in years 3 & 4 21

22. © Employee Benefit Research Institute 2016 HSA Reduced Rx Use: Number of Prescription Fills, by Group, 2006-2010 22 Source: Fronstin, et al. Health Affairs (June 2013).

23. © Employee Benefit Research Institute 2016 Proportion of Population With Various Chronic Conditions that Adherent to Medication 23 Source: Fronstin, et al. American Journal of Managed Care (Dec. 2013).

24. © Employee Benefit Research Institute 2016 Proportion of Population With Various Chronic Conditions that Adherent to Medication 24 Source: Fronstin, et al. American Journal of Managed Care (Dec. 2013).

25. © Employee Benefit Research Institute 2016 HSA Increased ER Use: Number of Emergency Room Visits (Per 1,000), by Group, 2006-2010 25 Source: Fronstin, et al. Health Affairs (June 2013).

26. © Employee Benefit Research Institute 2016 Summary of Findings – Quality Measures Preventive services not subject to deductible Office visits fell in years 1 and 4 Breast cancer screening fell in year 1, higher in year 4 Cervical cancer screening fell in years 1 and 4 Colorectal cancer screening lower in all years Other services Medication monitoring fell Higher use of imaging for back pain & antibiotics for bronchitis Lower LDL testing for diabetics No impact on HbA1c testing for diabetics 26

27. © Employee Benefit Research Institute 2016 Estimates of the Impact of the HSA Plan on Health Spending Per Person, by Year 27 Statistical significance denoted as follows: *** p<0.01; ** p<0.05; * p<0.10. Source: Fronstin, et al. EBRI Issue Brief (July 2013).

28. © Employee Benefit Research Institute 2016 Summary of Findings – Spending Spending lower in all 4 years relative to 2006 baseline 25% lower in year 1 6% lower in year 4 Year 1, spending reductions across the board Year 4, spending reductions limited to Rx and lab 28

29. © Employee Benefit Research Institute 2016 Estimates of the Impact of the HSA Plan on Health Spending, by Type of Service and Year Cost Measure20072010 Total %-25%***-6%* Inpatient-33%11% Emergency Dept.-17%*5% Outpatient-13%*-12% Office visit-14%***5% SA/MH-22%**0% Laboratory-36%***-19%* Pharmacy-32%***-20%*** 29 Statistical significance denoted as follows: *** p<0.01; ** p<0.05; * p<0.10. Source: Fronstin, et al. EBRI Issue Brief (July 2013).

30. © Employee Benefit Research Institute 2016 Estimates of the Impact of the HSA Plan on Pharmacy Spending, by 2006 Spending Quintile and Year 2006 Spending Quintile20072010 1 (lowest spending)-47%***-41%*** 2-44%***-25%* 3-40%***-21%* 4-42%***-30%*** 5 (highest spending)-24%***-5% 30 Statistical significance denoted as follows: *** p<0.01; ** p<0.05; * p<0.10. Source: Fronstin, et al. EBRI Issue Brief (July 2013).

31. © Employee Benefit Research Institute 2016 Estimates of the Impact of the HSA Plan on Total Spending, by 2006 Spending Quintile and Year 2006 Spending Quintile20072010 1 (lowest spending)-12%9% 2-14%-1% 3-29%***-15%** 4-31%***-9% 5 (highest spending)-15%*-1% 31 Statistical significance denoted as follows: *** p<0.01; ** p<0.05; * p<0.10. Source: Fronstin, et al. EBRI Issue Brief (July 2013).

32. IMPACT OF HSA-ELIGIBLE HEALTH PLAN AND INCOME ON HEALTH SERVICES UTILIZATION 32

33. © Employee Benefit Research Institute 2016 Background Although the topic of price elasticity of demand for healthcare has been well- studied, little work has examined income elasticity of demand for healthcare Maybe because datasets often capture changes in health services “prices” (i.e., patient cost-sharing), but within-person income variation is uncommon Multiple sources of endogeneity/confounding Perhaps also because income is not usually considered a policy lever If healthcare is a necessity, then its demand should be income inelastic If healthcare is free (e.g., zero patient cost-sharing), income shouldn’t affect demand except through opportunity cost of patient time Under high cost-sharing, income may have a more profound impact Therefore the undesirable effects of high deductibles (i.e., underutilization of high value care) may differ according to income 33

34. © Employee Benefit Research Institute 2016 Research Questions What is the impact of CDHP on health services utilization? Does the impact of CDHP on health services utilization vary by income? Are lower income employees more likely to skimp on high valued services? If so, what are the longer term implications for complications, and higher health spending? Does income affect healthcare use independent of coverage generosity? Can a better plan design be designed that addresses income disparities? 34

35. © Employee Benefit Research Institute 2016 Data We used enrollment and health insurance claims data from a large firm that also provided annual income information for all of its employees We examined 150,000 to 200,000 full-time active members and their dependents (age<65), enrolled for 2 to 6 full calendar years between 2009- 2014 resulting in an unbalanced panel dataset of over 800,000 observations Included were members in either a preferred provider organization (PPO) plan or an HSA-eligible health plan In 2013, the employer encouraged enrollment in the HSA plan via low premiums, which led to an increase from 3% to 23% percent of the sample, but enrollment was still voluntary (i.e., not full-replacement) 35

36. © Employee Benefit Research Institute 2016 Health Plan Enrollment, by Year, 2009-2014 36

37. © Employee Benefit Research Institute 2016 Distribution of Annual Earnings, 2014 37

38. © Employee Benefit Research Institute 2016 Impact of HSA-Eligible Health Plan on Office Visits, by Employee Income 38 Source: EBRI analysis of administrative claims data. *** p<0.01, ** p<0.05, * p<0.10

39. © Employee Benefit Research Institute 2016 Impact of HSA-Eligible Health Plan on Prescription Drug Fills, by Employee Income 39 Source: EBRI analysis of administrative claims data. *** p<0.01, ** p<0.05, * p<0.10

40. © Employee Benefit Research Institute 2016 Impact of HSA-Eligible Health Plan on Likelihood of Receiving Preventive Services, by Employee Income (Per 100 Individuals) 40 Source: EBRI analysis of administrative claims data. *** p<0.01, ** p<0.05, * p<0.10

41. © Employee Benefit Research Institute 2016 Impact of HSA-Eligible Health Plan on Likelihood of Receiving Preventive Cancer Screenings, by Employee Income 41 Source: EBRI analysis of administrative claims data. *** p<0.01, ** p<0.05, * p<0.10

42. © Employee Benefit Research Institute 2016 Impact of HSA-Eligible Health Plan on Number of Inpatient Admissions and Emergency Department Visits, by Employee Income 42 Source: EBRI analysis of administrative claims data. *** p<0.01, ** p<0.05, * p<0.10

43. © Employee Benefit Research Institute 2016 Other Findings Health care services unaffected by enrollment in HSA-eligible health plan, both overall and by worker income Inpatient hospital days Avoidable emergency department visits Pneumonia vaccinations HPV vaccinations HbA1c testing for individuals with diabetes 43

44. © Employee Benefit Research Institute 2016 Learning Effect & Implications When 2014 claims data were included in the analysis, the positive effects on emergency department use and inpatient admissions for individuals with income below $50,000 were no longer statistically significant. Employers could provide higher HSA contributions to lower income workers Education needed regarding preventive services 44

45. © Employee Benefit Research Institute 2016 Unanswered Research Question The role of the account As account balances increase: Impact on use of health care services Impact on choice of deductible Who uses the account to save for retirement 45

46. EBRI : Just the Facts™ www.ebri.org www.choosetosave.org